使用node.js抓取有路网图书信息(原创)

之前写过使用python抓取有路网图书信息,见http://www.cnblogs.com/dyf6372/p/3529703.html。

最近想学习一下Node.js,所以想试试手,比较一下http抓取上的性能,采用事件驱动的Node.js比python好一些,以下上代码(刚学还未优化):

var http = require('http');
var iconv = require('iconv-lite');
var url = require('./gb2312_url_encode.js');

function getHtmlOptions(path){
    return  {
        hostname : 'www.youlu.net',
        port : 80,
        path : path,
        method : 'GET',
        headers : {
            'User-Agent' : 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/32.0.1700.77 Safari/537.36'
        }
    }
}

function getAllNumber(str){
    var re = /共有图书数量\r\n\s*[0-9]*/;
    var n_str = str.match(re)[0];
    re = /[0-9]{1,}/;
    return n_str.match(re)[0];
}

function getDetailList(str){
    var re = /\"\/\d+.*"/g;
    var n_array = str.match(re);
    //console.log(n_array);
    var result_array = [];
    if(n_array == null){
        return [];
    }
    for(var i=0;i<n_array.length;i++){
        var tmp = n_array[i];
        re = /\d+/;
        var tmp_item = tmp.match(re);
        //console.log(tmp_item[0]);
        if(result_array.indexOf(tmp_item[0])<0){
            re = /alt=\".*\"/;
            var n_tmp_item = tmp.match(re);
            if(n_tmp_item == null) continue;
            var book_name = n_tmp_item[0].substring(5);
            book_name = book_name.substring(0,book_name.length-1);
            result_array.push(tmp_item[0]);
            searchDetail(tmp_item[0],book_name);
        }
    }
    return result_array;
}

function getRealBookNumber(str,book_name,url){
    re = /startRequestBookBuyLink(.*)/;
    n_array = str.match(re);
    var num = n_array[0].split(',')[3];
    num = num.substring(2,num.length-1);
    if(num > 0){
        console.log("------------------------------------");
        console.log(book_name);
        console.log("数目:"+num+" url:"+url);
    }
}

function firstSearch(query_book_name){
    var options = getHtmlOptions('/search/result/default.aspx?isbn=&publisherName=&author=&bookName='+query_book_name);
    var req = http.request(options,function(res){
        if(res.statusCode != 200){
            console.log("请求/search/result/default.aspx?isbn=&publisherName=&author=&bookName="+query_book_name+"发生错误了:请求返回非200,返回码:"+res.statusCode+"\n");
            return;
        }

        var chunks = [];
        var size = 0;

        res.on('data',function(chunk){
            chunks.push(chunk);
            size += chunk.length;
        });

        res.on('end',function(){
            var returnMsg = Buffer.concat(chunks,size);
            var nreturnMsg = iconv.decode(returnMsg,'gb2312');
            var allNumber = getAllNumber(nreturnMsg);
            if(allNumber>0){
                var allPage = (allNumber/20+0.5).toFixed(0);
                console.log("该图书数目为:"+allNumber);
                console.log("共有页数:"+allPage);
                for(var i = 1; i<=allPage; i++){
                    searchEachPage(query_book_name,i,allNumber);
                }
            }else{
                console.log("该图书数目为0."+"\n");
            }
        });
    });

    req.on('error', function(e) {
        console.log('problem with request: ' + e.message);
    });

    req.end();
}

function searchEachPage(query_book_name,pageIndex,rowCount){
    var options = getHtmlOptions("/search/result/?isbn=&publisherName=&author=&bookName="+query_book_name+"&pageIndex="+pageIndex+"&rowCount="+rowCount+"&searchIn=");
    var req = http.request(options,function(res){
        if(res.statusCode != 200){
            console.log("请求"+"/search/result/?isbn=&publisherName=&author=&bookName="+query_book_name+"&pageIndex="+pageIndex+"&rowCount="+rowCount+"&searchIn="+"发生错误了:请求返回非200,返回码:"+res.statusCode+"\n");
            return;
        }

        var chunks = [];
        var size = 0;

        res.on('data',function(chunk){
            chunks.push(chunk);
            size += chunk.length;
        });

        res.on('end',function(){
            var returnMsg = Buffer.concat(chunks,size);
            var nreturnMsg = iconv.decode(returnMsg,'gb2312');
            getDetailList(nreturnMsg);
        });
    });

    req.on('error', function(e) {
        console.log('problem with request: ' + e.message);
    });

    req.end();
}


function searchDetail(detail_number,book_name){
    var options = getHtmlOptions("/"+detail_number);
    var req = http.request(options,function(res){
        if(res.statusCode != 200){
            console.log("请求"+"/"+detail_number+"发生错误了:请求返回非200,返回码:"+res.statusCode+"\n");
            return;
        }

        var chunks = [];
        var size = 0;

        res.on('data',function(chunk){
            chunks.push(chunk);
            size += chunk.length;
        });

        res.on('end',function(){
            var returnMsg = Buffer.concat(chunks,size);
            var nreturnMsg = iconv.decode(returnMsg,'gb2312');
            getRealBookNumber(nreturnMsg,book_name,"http://www.youlu.net/"+detail_number);
        });
    });

    req.on('error', function(e) {
        console.log('problem with request: ' + e.message);
    });

    req.end();
}

var query_book_name= url.URLEncode('java');
firstSearch(query_book_name);

你可能感兴趣的:(node.js)